Setting Up AI Providers

FloWords supports multiple AI providers for text enhancement. Choose based on your privacy needs, budget, and desired features.

---

Provider Comparison

Provider	Privacy	Cost	Speed	Quality
Ollama	Local/Best	Free	Variable	Good-Excellent
OpenAI	Cloud	$$$	Fast	Excellent
Anthropic	Cloud	$$$	Fast	Excellent

Tip

For the best balance of privacy and quality, try Ollama with a local model. It’s free and keeps your data on your Mac.

---

OpenAI Setup

OpenAI’s GPT models offer reliable, high-quality enhancement.

Available Models

Model	Speed	Quality	Cost
GPT-4o	Fast	Best	$$$
GPT-4o-mini	Fastest	Great	$
GPT-4	Medium	Excellent	$$$$
GPT-3.5-turbo	Fast	Good	$

Configuration Steps

1. Get API Key

   • Go to platform.openai.com
   • Sign in or create account
   • Navigate to API Keys
   • Create new secret key
   • Copy the key (shown only once)

2. Configure FloWords

   • Open Settings > Enhancement
   • Enable AI Enhancement
   • Select OpenAI as provider
   • Paste your API key
   • Select your preferred model

3. Test Connection

   • Click Test button
   • Should show “Connection successful”
   • Make a test recording

Cost Management

OpenAI charges per token (roughly per word):

Model	Input Cost	Output Cost
GPT-4o-mini	$0.15/1M	$0.60/1M
GPT-4o	$2.50/1M	$10.00/1M
GPT-4	$30.00/1M	$60.00/1M

Note

GPT-4o-mini is recommended for most users. Excellent quality at a fraction of the cost.

---

Anthropic Setup

Anthropic’s Claude models excel at nuanced, safe text processing.

Available Models

Model	Speed	Quality	Cost
Claude 3.5 Sonnet	Fast	Excellent	$$
Claude 3 Opus	Medium	Best	$$$$
Claude 3 Haiku	Fastest	Good	$

Configuration Steps

1. Get API Key

   • Go to console.anthropic.com
   • Sign in or create account
   • Navigate to API Keys
   • Create new key
   • Copy the key

2. Configure FloWords

   • Open Settings > Enhancement
   • Enable AI Enhancement
   • Select Anthropic as provider
   • Paste your API key
   • Select your preferred model

3. Test Connection

   • Click Test button
   • Verify successful connection
   • Make a test recording

Why Choose Anthropic

• Excellent at following instructions
• Strong safety and content filtering
• Great for professional content
• Nuanced writing capabilities

---

Ollama Setup (Local AI)

Ollama runs AI models locally on your Mac. Maximum privacy, no API costs.

Requirements

• macOS 14.0 (Sonoma) or later
• Apple Silicon recommended (M1/M2/M3)
• 8GB+ RAM (16GB+ for larger models)
• ~4-10GB disk space per model

Installation Steps

1. Install Ollama

   • Go to ollama.com
   • Download for macOS
   • Run the installer
   • Ollama runs in your menu bar

2. Download a Model

   Open Terminal and run:

   # Recommended for most users
   ollama pull llama3.2
   
   # Alternative options
   ollama pull mistral
   ollama pull phi3
   ollama pull gemma2

3. Configure FloWords

   • Open Settings > Enhancement
   • Enable AI Enhancement
   • Select Ollama as provider
   • Select your downloaded model
   • Test the connection

4. Verify Setup

   • Ollama should be running (check menu bar)
   • Make a test recording
   • Check enhancement works

Recommended Local Models

Model	Size	RAM Needed	Quality	Speed
Llama 3.2 3B	2 GB	4 GB	Good	Fast
Llama 3.2 8B	4.7 GB	8 GB	Great	Medium
Mistral 7B	4.1 GB	8 GB	Great	Medium
Phi-3 mini	2.3 GB	4 GB	Good	Fast
Gemma 2 9B	5.4 GB	10 GB	Excellent	Slower

Tip

Start with Llama 3.2 3B for a good balance of quality and speed. Upgrade to larger models if you need better results.

Ollama Commands

Common commands in Terminal:

ollama list

ollama pull llama3.2

ollama rm model-name

ollama serve

ollama run llama3.2 "Hello, test"

---

Provider Switching

You can easily switch between providers:

1. Go to Settings > Enhancement
2. Select different provider
3. Enter credentials if needed
4. Save settings

Your API keys are stored securely and you can switch anytime.

---

Troubleshooting

OpenAI Issues

“Invalid API Key”

• Check for typos in the key
• Ensure key hasn’t been revoked
• Generate a new key if needed

“Rate Limit Exceeded”

• Wait a few minutes
• Check your OpenAI usage limits
• Consider upgrading your plan

“Insufficient Credits”

• Add credits at platform.openai.com
• Set up billing

Anthropic Issues

“Authentication Failed”

• Verify API key is correct
• Check account status
• Regenerate key if needed

“Context Too Long”

• Use shorter recordings
• Select a model with larger context

Ollama Issues

“Connection Refused”

• Ensure Ollama is running
• Check for Ollama in menu bar
• Restart Ollama: killall ollama && ollama serve

“Model Not Found”

• Run ollama pull model-name
• Check model name spelling
• Run ollama list to see installed models

“Out of Memory”

• Use a smaller model
• Close other applications
• Check Activity Monitor

---

Security Best Practices

API Keys

• Never share your API keys
• Don’t commit keys to version control
• Rotate keys periodically
• Use separate keys for different apps

Data Privacy

• Cloud providers process your text on their servers
• Review provider privacy policies
• Use Ollama for sensitive content
• Consider what data you’re enhancing

Cost Protection

• Set spending limits in provider dashboards
• Monitor usage regularly
• Use cost-effective models for routine tasks
• Reserve premium models for important content

---

Next Steps

• Configure Enhancement Prompts to customize how your text is processed
• Enable Context Awareness for smarter enhancements